1. Field of the Invention
The invention presented is a procedure for the targeted recognition and characterization of defects on the surface of magnetic disks. In particular, the invention recognizes those defects which are in the submicrometer range. It is also a device for executing such a procedure.
2. Description of Related Art
Magnetic disks are data storage devices with a very large storage capacity. Currently 1 GB can be stored on a magnetic disk of about 65 mm in size. The disks must have both exact magnetic and mechanical as well as certain tribological characteristics. When later used in a disk drive, the disks are rotated by a spindle motor at speeds typically ranging from 3600 to 10000 revolutions per minute. This means that the outer edge of the disk reaches a speed of up to 100 km/h while the write/read head is flying at less than one twenty-thousandths of a millimeter (&lt;50 nm) away from the disk surface. These quality requirements can only be achieved through the greatest precision in manufacture and statistical process controls.
To achieve higher storage densities, a thin magnetic film medium is used for recording data on modern high-capacity magnetic disks. Defects which are distributed across the surface of these thin film disks can significantly worsen the production yield of disk drives. These defects can be caused by a local change in the topography or the magnetic structure. Examples of this are the so-called "thermal asperities" or "missing bits". Thus submicrometer defects, i.e., defects &lt;=1 .mu.m, can lead to defective magnetic disks. The specific location and the exact characterization of these defects is therefore extremely important for manufacturing magnetic disks.
It is particularly important to create a correlation between the submicrometer defects, the topography of the magnetic disk or its magnetic structure and the resulting errors in read signals. This is the basis for a general analysis of defects on the disk and the assessment of its relevance for error-free function of the hard-disk drive in which these magnetic disks are used.
The tests used to locate errors in the read signal, cannot characterize the cause of the error. This characterization with regard to topography and the magnetic characteristics may be carried out using an atomic force microscope (AFM), magnetic force microscope (MFM) or other appropriate sensing device. The maximum area which can be recorded with such devices is typically 100 .mu.m.times.100 .mu.m.
The determination of the cause of an error as a magnetic and/or topographic defect is possible with the current state of the art using the steps of a) magnetically marking the defect on the magnetic (MAGNETIC) tester, b) highlighting the mark using ferrofluid, c) then mechanical marking the position, d) removal of the ferrofluid, and e) subsequent examination, e.g. by AFM and/or MFM. In this procedure it is possible that on removing the ferrofluid relevant defects are also removed and the disk may be soiled or otherwise damaged. In most cases, this makes specific analysis impossible. In addition, the average time requirement for this procedure starts at 1 day or more for one single defect, making it impractical for production use.
There is a further possibility of improving the magnetic marking technique as described in the following paragraph. In this, the disadvantages in using ferrofluid stated above are avoided. A direct allocation of signal faults and defects is possible which can give important information on the cause of the fault. However, even with this method, the time requirement is still high (&gt;=1 day). Also, an examination of all relevant cases has to be excluded in this procedure and the analysis limited to test samples.
The magnetic disk along with its defects is first inserted into a MAG tester. Defects are detected at typical rotational speeds as an error in the read signal (see FIG. 3, upper curve) and its coordinates (radius r.sub.i, angle position .phi..sub.i) is established. The angle .phi. is counted from a prestated reference position (Index 0) given by the spindle drive of the magnetic disk. The radial position r.sub.i is measured outwards from the rotation axis of the spindle. The programs required for this are available to the tester. Then a specific defect is selected. On the radius r.sub.i of the magnetic disk, a track is now written, allowing a 50-100 .mu.m hole around it so that the defect itself is not overwritten. The interruption of the recorded data on the track now marks the position of the defect. A second track is written at r.sub.i -10 .mu.m which only extends from index 0 to .phi..sub.i
If more than one defect is examined, then the previously described steps must be repeated for each additional defect. As the marking does take up some space, it is not possible to simultaneously mark defects whose radial coordinates are narrower than the width of the marking.
Then a mark for index 0 is applied mechanically to the magnetic disk. The width of this mark is generally &gt;500 .mu.m. The magnetic disk is now removed from the MAG tester and inserted into the MFM (or AFM). The position of the source of the coordinates is lost in this removal and reinsertion. The offset to the new source can be up to 700 .mu.m.
In the next step, the radius r.sub.i and the angle .phi..sub.i are roughly set by hand and the MFM measurement is started. Should one of the previously written tracks be visible within the measuring window, then the measuring window is centered on it. For this, a sample table procedure is required. If no track can be found, then the table is moved radially and a new measurement started until the tracks are found.
After the radial coordinates are found, the angle coordinates are now set, for which MFM or AFM measurements are again required, before the actual MFM or AFM measurement can take place (see FIG. 3, lower curve for AFM or FIG. 4 for MFM). To locate the exact position (r.sub.i, .phi..sub.i), a practiced operator requires up to 180 minutes. The total time requirement for this method is therefore approximately 1 day or more per defect.
The current procedures require at least two devices to locate the defect position, to mark it and to characterize it. In transferring the disks from the first device on which the detection of the defects was made using the read signal, to the second device which carries out the actual characterization, the disk must be removed from the spindle holder. In doing so, the reference points of the spindle coordinate system are lost. A reconstruction of the coordinate system using the described marking technique is very laborious and of only limited accuracy as the magnetic disk cannot be stored centrally either on the spin stand or in AFM/MFM because of the normal production tolerances.